The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources including HubSpot contact records, Apollo match results, and Clay enrichment data.
Ingestion points: Contact data is pulled via execute_sql in Stage 1, while enrichment data enters the context in Stages 4 and 5.
Boundary markers: There are no explicit delimiters or instructions to ignore embedded commands when processing or reporting the external contact strings.
Capability inventory: The skill possesses capabilities to execute SQL queries, write back to the HubSpot CRM, send Slack messages, and write files to the local skill-state directory.
Sanitization: The instructions do not specify sanitization for lead-provided strings (such as job titles or names) before they are interpolated into reports or Slack notifications.
[COMMAND_EXECUTION]: Uses the execute_sql tool to query the HubSpot CRM database.
The SQL logic is hardcoded within the skill workflow rather than being dynamically generated from user input, which limits the risk of SQL injection attacks.
[DATA_EXFILTRATION]: Transfers HubSpot PII, including names and email addresses, to external services for enrichment.
This behavior is the primary functional purpose of the skill and uses well-known technology services (Apollo and Clay).
The data transfer is restricted to the specific lead fields required for enrichment.

prospect-enrich-skill